Kinetic Mechanisms of the Oxygenase from a Two-component Enzyme, p-Hydroxyphenylacetate 3-Hydroxylase from Acinetobacter baumannii*

p-Hydroxyphenylacetate hydroxylase (HPAH) from Acinetobacter baumannii catalyzes the hydroxylation of p-hydroxyphenylacetate (HPA) to form 3,4-dihydroxyphenylacetate (DHPA). The enzyme system is composed of two proteins: an FMN reductase (C1) and an oxygenase that uses FMNH– (C2). We report detailed transient kinetics studies at 4 °C of the reaction mechanism of C2.C2 binds rapidly and tightly to reduced FMN (Kd, 1.2 ± 0.2 μm), but less tightly to oxidized FMN (Kd, 250 ± 50 μm). The complex of C -FMNH–2 reacted with oxygen to form C(4a)-hydroperoxy-FMN at 1.1 ± 0.1 × 106 m–1 s–1, whereas the C -FMNH–2 -HPA complex reacted with oxygen to form C(4a)-hydroperoxy-FMN-HPA more slowly (k = 4.8 ± 0.2 × 104 m–1 s–1). The kinetic mechanism of C2 was shown to be a preferential random order type, in which HPA or oxygen can initially bind to the C -FMNH–2 complex, but the preferred path was oxygen reacting with C -FMNH–2 to form the C(4a)-hydroperoxy-FMN intermediate prior to HPA binding. Hydroxylation occurs from the ternary complex with a rate constant of 20 s–1 to form the C2-C(4a)-hydroxy-FMN-DHPA complex. At high HPA concentrations (>0.5 mm), HPA formed a dead end complex with the C2-C(4a)-hydroxy-FMN intermediate (similar to single component flavoprotein hydroxylases), thus inhibiting the bound flavin from returning to the oxidized form. When FADH– was used, C(4a)-hydroperoxy-FAD, C(4a)-hydroxy-FAD, and product were formed at rates similar to those with FMNH–. Thus, C2 has the unusual ability to use both common flavin cofactors in catalysis.

Hydroxylation of p-hydroxyphenylacetate (HPA) to form 3,4-dihydroxyphenylacetate (DHPA) by HPAH is especially interesting because the same reaction is carried out by at least three types of two-component enzymes. The first HPAH purified was from P. putida, and it was shown to have FAD tightly bound to the smaller protein, and the larger protein (at that time) was thought to be a coupling protein enabling hydroxylation (4,16). A different HPAH system was later isolated from E. coli W, and studies have shown that the smaller component (HpaC) is a flavin reductase that generates reduced FAD to be transferred to the larger component (HpaB) to hydroxylate HPA (5,17). A detailed analysis of the mechanism of the E. coli-type HPAH is now in progress using the homologue from P. aeruginosa (18). The oxygenase in this system exhibits complex dynamics in catalysis (19).
Our group has isolated HPAH from A. baumannii and shown that the enzyme is quite different from the analogous HPAH enzymes from either P. putida or E. coli (6,15,20). The A. baumannii HPAH is a two protein enzyme system consisting of a smaller reductase component (C 1 ) and a larger oxygenase component (C 2 ) (6). Sequence and several catalytic properties indicate that both components are different from others in the two protein class of aromatic hydroxylases (15,20). Our recent investigations of the reaction mechanisms of C 1 have shown that HPA controls the reduction of C 1 -bound FMN by NADH by shifting the enzyme into a more active conformation (20). By contrast, HPA has no effect at all on the activity of the reductase from the E. coli-type HPAH from P. aeruginosa (18). The HPAH from P. putida (above) (16) requires fresh examination based upon our current knowledge. It is possible that this enzyme system operates in a manner similar to the system from A. baumannii, but the essential experiments to test this possibility have not been carried out. C 2 shows little sequence similarity to other oxygenases in the same class, and is unique for its ability to use reduced forms of riboflavin, FMN, or FAD to catalyze hydroxylations (6,15). The overall reaction of C 2 is described in Fig. 1. When C 2 was mixed with reduced flavin and a limited amount of oxygen, an intermediate spectrum resembling that of a C(4a)-oxygen adduct of flavin was observed (6). Similar observations were made in the analogous reactions of the oxygenase component involved in biosynthesis of actinorhodin in Streptomyces coelicolor (ActVA) (21,22) and with chlorophenol 4-monooxygenase (9). Despite preliminary observations that C(4a)-oxygenated intermediates are likely to be involved in oxygenation reactions of these oxygenase components, investigations by presteady state methods to elucidate the enzyme reaction mechanism in detail have never been carried out. In this article, we report investigations on the reaction of oxygen with C 2 and reduced flavin using single mixing and double mixing stopped-flow spectrophotometry. The results comprehensively elucidate the reaction mechanism of C 2 , the order of substrate binding, and the binding constants of FMNH Ϫ and HPA to the enzyme.

MATERIALS AND METHODS
Reagents-NAD ϩ , NADH, FAD, glucose, and glucose oxidase were from Sigma. FMN was prepared by conversion of FAD to FMN with snake venom from Crotalus adamanteus (23). In brief, FAD (2.5 mg/ml) and venom (50 g/ml) in 20 mM potassium phosphate buffer, pH 7.0 were incubated overnight in the dark. The reaction mixture was loaded onto a C 18 Sep-Pak cartridge (Waters), previously equilibrated with 20 mM potassium phosphate buffer, pH 7.0, and the cartridge was washed with 10 mM potassium phosphate buffer, pH 7.0. FMN was eluted with water, and the solution was freeze-dried. Concentrations of the following compounds were determined using known extinction coefficients at pH 7.0: NADH, ⑀ 340 ϭ 6.22 mM Ϫ1 cm Ϫ1 ; FAD, ⑀ 4500 ϭ 11.3 mM Ϫ1 cm Ϫ1 ; FMN, ⑀ 446 ϭ 12.2 mM Ϫ1 cm Ϫ1 and HPA, ⑀ 277 ϭ 1.5 mM Ϫ1 cm Ϫ1 (6). C 2 used in this study was cloned, expressed, and prepared as previously described (15). The concentration of C 2 was estimated from the extinction coefficient (based on amino acid sequence) of ⑀ 280 ϭ 56.7 mM Ϫ1 cm Ϫ1 .
Enzyme Activity-Enzyme hydroxylation activity was detected in real time using a coupling reaction involving 3,4-dihydroxyphenylacetate dioxygenase (DHPAO) to convert the DHPA product of C 2 to 5-carboxymethyl-2-hydroxy-muconate semi-aldehyde (CHS). This yellow compound has a maximum absorbance at 380 nm that is dependent upon pH (4,6).
Spectroscopic Studies-UV-visible absorbance spectra were recorded with a Hewlett Packard diode array spectrophotometer (HP 8453A), or a Shimadzu 2501PC spectrophotometer. Fluorescence measurements were carried out with a Shimadzu RF5301PC spectrofluorometer. All these instruments were equipped with thermostatic cell compartments.
Determination of the K d for Binding Oxidized FMN to C 2 -The measurements were performed by an ultrafiltration method using Centriprep Y-M 10 from Amicon. Solutions were composed of 10 M FMN in 50 mM sodium phosphate buffer, pH 7.0, and various concentrations of C 2 , (20,40,80,160, and 200 M) in a 10-ml total volume. Each solution was centrifuged at 3,200 rpm, 4°C for 15 min to obtain a filtrate of ϳ1 ml (to minimize change in volume). The filtrate and retentate were analyzed for the amount of free and bound FMN, respectively. Ratios of the free and bound species were used to calculate the K d value.
Rapid Reaction Experiments-Reactions were carried out in 50 mM sodium phosphate buffer, pH 7.0, 4°C, unless otherwise specified. Rapid kinetics measurements were performed with a Hi-Tech Scientific Model SF-61DX in double mixing mode, or with either a model SF-61SX or a SF-61DX stopped-flow spectrophotometer in single mixing mode. The optical pathlengths of the observation cells were 1 cm. The stopped-flow apparatus was made anaerobic by flushing the flow system with an oxygen-scrubbing solution consisting of 400 M glucose, 1 mg/ml glucose oxidase (15.5 unit/ml), and 4.8 g/ml catalase in 50 mM sodium phosphate buffer, pH 7.0. The oxygen-scrubbing solution was allowed to stand in the flow system overnight and was then thoroughly rinsed with anaerobic buffer before experiments.
To study the oxidative half-reaction of C 2 , enzyme, or enzyme plus substrate and oxidized FMN in 50 mM sodium phosphate buffer, pH 7.0, were made anaerobic in glass tonometers by several cycles of evacuation followed by equilibration with argon that had been passed through an Oxyclear oxygen removal column (Labclear). Enzyme was anaerobically reduced with a solution of sodium dithionite (ϳ5 mg/ml in 100 mM potassium phosphate buffer, pH 7.0) delivered from a syringe attached to the tonometer, and the reduction was monitored by UV-visible spectrophotometry. Solutions with various concentrations of oxygen were prepared by equilibrating buffer with air or with certified oxygen/nitrogen gas mixtures. Determinations of rate constants were obtained by fitting plots of apparent rate constants (k obs ) versus concentration of substrate with a Marquardt-Levenberg non-linear fit algorithm that is included in the KaleidaGraph software (Synergy Software). The k obs from kinetic traces were calculated from exponential fits using KinetA-syst3 software (Hi-Tech Scientific, Salisbury, UK) or program A (written at the University of Michigan by Rong Chang, Jung-yen Chiu, Joel Dinverno, and D. P. Ballou).

RESULTS
Reaction of C 2 -FMNH Ϫ with O 2 in the Absence of HPA-A solution of FMN (16 M) plus C 2 (25 M) was placed in a glass tonometer equipped with a quartz cuvette, and made anaerobic as described under "Materials and Methods." An anaerobic solution of sodium dithionite was delivered into the tonometer to stoichiometrically reduce the FMN (see "Materials and Methods"). The resulting C 2 -FMNH Ϫ complex was loaded onto the stopped-flow spectrophotometer, where its reaction with oxygen was monitored at 380 and 446 nm (Fig. 2). A significant fraction of the first phase, which is shown by an increase in absorbance at 380 nm and no change at 446 nm, occurred during the dead time of the instrument (ϳ0.002 s) and was complete by 0.006 -0.02 s (high to low oxygen concentration, Fig. 2). The plot of k obs of this phase versus oxygen concentration was linear, yielding a second-order rate constant of 1.1 Ϯ 0.1 ϫ 10 6 M Ϫ1 s Ϫ1 (inset in Fig. 2). When absorbance values at the end of the first phase (reaction time of 0.01 s of highest oxygen reaction) at various wavelengths were plotted, the spectrum B in Fig. 3 was obtained. This spectrum has characteristics typical for a flavin-C(4a)-adduct with maximum absorbance at 380 nm. Based upon analogy to the reactions catalyzed by one component hydroxylases and the condition that HPA is absent, the spectrum B in Fig. 3 is likely to be C(4a)-hydroperoxy-FMN (1-3). Spectrum B is also similar to that of C(4a)-hydroperoxyflavins generally found in the class of single component aromatic hydroxylases (1, 3, 24 -26), as well as for luciferase (the first two-component flavin-dependent oxygenase studied in detail) (27,28), cyclohexanone monooxygenase (29) Fig. 2). A small increase in absorbance at 380 nm was also observed between 0.01 and 1 s, and the k obs describing this phase was also dependent on oxygen concentration. Based on the K d value of C 2 -FMNH Ϫ of 1.2 Ϯ 0.2 M (described in the next paragraph), ϳ1.6 M free FMNH Ϫ is present under these reaction conditions. Therefore, this small absorbance change is likely to be because of free FMNH Ϫ reacting with oxygen.
When the same reaction of the C 2 -FMNH Ϫ complex was investigated with fluorescence detection using the excitation wavelengths of 360 -446 nm and emission at wavelengths of greater than 500 nm, fluorescence increases were only observed with formation of the final species, oxidized FMN (data not shown). This result indicated that the C 2 -C(4a)-hydroperoxy-FMN was non-fluorescent.
Determination of Binding Constants of Reduced and Oxidized Flavin to C 2 -Free reduced FMN was mixed with air-saturated C 2 solution in the stopped-flow apparatus, resulting in a reaction with kinetic traces nearly identical to those obtained when preformed C 2 -FMNH Ϫ was mixed with oxygen, as shown in Fig. 2. This result implies that binding of FMNH Ϫ to C 2 is much faster than the oxidation of free FMNH Ϫ by oxygen (32), and also greater than the rate of formation of the C(4a)flavin hydroperoxide at 0.13 mM oxygen, 185 Ϯ 9 s Ϫ1 (Fig. 2). Thus, the rate constant for C 2 binding to FMNH Ϫ is likely to be Ն10 7 M Ϫ1 s Ϫ1 (k 1 in Fig. 10).
Therefore, when FMNH Ϫ (16 M) was mixed with various concentrations of C 2 in air-saturated buffer in the stopped-flow spectrophotometer, the absorbance increase at 380 nm during the first phase ( Fig.  4), because of the C(4a)-hydroperoxy FMN formed, was also directly dependent on the amount of C 2 -FMNH Ϫ complex initially present. In the absence of C 2 (the lowest trace), the absorbance increased with a t1 ⁄ 2 ϳ0.7 s as free FMNH Ϫ oxidized to FMN in a complex autocatalytic reaction (32). As the concentration of C 2 increased, less auto-oxidation of FMNH Ϫ is observed. Therefore, the increase in absorbance observed at 0.04 s represents the amount of C 2 -FMNH Ϫ present at the start of the reaction, and the plot of this change in absorbance versus the free C 2 concentration represents the binding isotherm for FMNH Ϫ . The plot (inset in Fig. 4) shows that the absorbance increase is hyperbolically dependent on C 2 concentration. A K d (referred to as K d A in the kinetic scheme in Fig. 10) value for the complex was calculated to be   The K d for the binding of oxidized FMN to C 2 was determined by an ultrafiltration method described under "Materials and Methods." This experiment indicated that the K d of C 2 -FMN was 250 Ϯ 50 M. The large uncertainty occurred because values could not be obtained at appropriately high C 2 concentrations.
Reaction of the C 2 -FMNH Ϫ -HPA Complex with Oxygen-The reaction of C 2 in the presence of substrate was investigated by mixing an anaerobic solution of FMNH Ϫ (16 M), C 2 (25 M), and 2 mM HPA with buffer containing various oxygen concentrations in the stopped-flow spectrophotometer (Fig. 5). The rate of formation of C 2 intermediate in presence of HPA was second order with respect to oxygen; however, the reaction is considerably slower than when no HPA is present (compare the increases at 380 nm in Fig. 2 to those in Fig. 5). An obvious interpretation of this result is that the reaction with oxygen is slower when HPA is bound to the enzyme. This interpretation was later verified (Figs. 7 and 8). In experiments where the concentration of HPA was varied in double mixing stopped-flow experiments, the rate of formation of C(4a)-hydroperoxy-FMN decreased with increasing concentrations of HPA (data not shown). Fig. 5 shows that the reaction monitored at 380 nm consists of 4 phases. For example, with 130 M oxygen (concentration after mixing), the first phase consisted of an increase of absorbance of ϳ0.026 AU occurring during the dead time, and continuing until ϳ0.012 s. With the highest oxygen concentration used (1.03 mM), this phase was complete during the dead time of the instrument. This first phase was dependent on oxygen concentration and is characterized by a rate constant of 1.1 Ϯ 0.1 ϫ 10 6 M Ϫ1 s Ϫ1 (data not shown). This value is the same as that observed in the reaction of C 2 -FMNH Ϫ with oxygen in Fig. 2 (with no HPA present). The second phase (ϳ0.012-0.6 s) of ϳ0.079 AU at 380 nm was also dependent on oxygen concentration and was characterized by a second order rate constant of 4.8 Ϯ 0.2 ϫ 10 4 M Ϫ1 s Ϫ1 (data not shown). Therefore, the first phase is likely to be the reaction of oxygen with C 2 -FMNH Ϫ without HPA bound, whereas the second phase is due to the ternary complex C 2 -FMNH Ϫ -HPA reacting with oxygen. The third phase is a lag in absorbance at 380 nm and corresponds to a process with a rate of ϳ17-22 s Ϫ1 (k 6 in Fig. 10). This phase can only be resolved clearly in the reaction with 1.03 mM oxygen. The decrease in absorbance because of the fourth phase (0.7-20 s) was coupled with a large increase at 446 nm; this phase was fitted with a rate constant (0.35 Ϯ 0.02 s Ϫ1 ) that was independent of oxygen concentration.
The absorption spectra of intermediates were calculated from the traces over a range of wavelengths of the reaction with 1.03 mM oxygen using the following rate constants: 54 s Ϫ1 for formation of the first intermediate, 20 s Ϫ1 for the second intermediate, and 0.35 s Ϫ1 for the final step in which oxidized FMN is formed. The analysis shows that spectra of the two intermediates are very similar (inset of Fig. 5), and have absorption characteristics similar to C(4a)-intermediates found for the single component flavoprotein hydroxylases (1, 24 -26, 33). This also implies that the first and second intermediates are likely to be C 2 -C(4a)-hydroperoxy-FMN-HPA complex and C 2 -C(4a)-hydroxy-FMN-product complex, respectively (inset of Fig. 5). The slight increased absorbance in the region of 450 nm of the second intermediate in the inset to Fig. 5 is unlikely to belong to absorption of C 2 -C(4a)hydroxy-FMN, but rather to a small amount of oxidized FMN resulting from an uncoupling pathway that does not result in hydroxylation (2, 24 -26).
Therefore, we conclude that the first phase is the reaction of C 2 -FMNH Ϫ (without HPA bound) whereas the second phase is the formation of C 2 -C(4a)-hydroperoxy-FMN-HPA complex. This also implies that the binding of HPA to the enzyme decreases the rate of formation of C 2 -C(4a)-hydroperoxy-FMN about 20-fold. We interpret the third phase to be the hydroxylation step where C 2 -C(4a)-hydroperoxy-FMN reacted with HPA to form the C 2 -C(4a)-hydroxy-FMN and DHPA. The C(4a)-hydroperoxy-FMN and C(4a)-hydroxy-FMN species have very similar spectra with this enzyme (see below), causing the absorbance change upon hydroxylation to be very small. Because of this small absorbance change, the rate constant for the hydroxylation step could not be determined accurately by this procedure. The fourth phase was caused by the dehydration of C 2 -C(4a)-hydroxy-FMN to yield the oxidized FMN species.
To verify further if the C 2 -FMNH Ϫ -HPA complex has indeed led to hydroxylation as described above, DHPA product formed under the conditions used in stopped-flow experiments was determined. Reaction samples were collected from the stopped-flow instrument and quantified by HPLC methods. Solutions of FMNH Ϫ (50 M), C 2 (80 M), and HPA (2 mM) were mixed with air-saturated buffer containing 2 mM HPA at 4°C in the stopped-flow spectrophotometer. The reaction solutions from several shots were collected for analysis. The collected solutions were ultrafiltered using Centricons to remove the enzyme, and the samples were analyzed for DHPA by HPLC methods described previously (6). The analysis showed that 73 Ϯ 4% of HPA was hydroxylated to form the DHPA product from the ternary complex under these conditions (Table 1).
HPA Is Not the First Substrate Binding to C 2 -The reaction of C 2 involves three substrates (Fig. 1), HPA, FMNH Ϫ , and oxygen. In this section we describe experiments to determine the sequence of binding of these compounds to C 2 . Fig. 6 shows experiments of the reaction to form the C 2 -C(4a)-hydroperoxy-FMN involving various premixing protocols. Trace A shows the reaction of the C 2 -FMNH Ϫ -HPA complex with O 2 (from Fig. 5), and trace B shows the reaction of the C 2 -FMNH Ϫ complex with O 2 and HPA. These demonstrate that the C 2 -FMNH Ϫ -HPA complex reacts with O 2 much more slowly than does the C 2 -FMNH Ϫ complex. We used this information to examine whether C 2 can effectively bind HPA in the absence of FMNH Ϫ . If such a complex does form, it would be expected that this complex in the presence of O 2 would react with FMNH Ϫ to form C 2 -C(4a)-hydroperoxy-FMN at the slower rate, as seen in trace A. Upon mixing a solution containing C 2 , HPA, and oxygen with FMNH Ϫ in the stopped-flow spectrophotometer (trace D, Fig. 6), C 2 -C(4a)-hydroperoxy-FMN formed at the same rate as the reaction with free C 2 -FMNH Ϫ (trace B) as well as that with mixing C 2 in air-saturated solution with FMNH Ϫ (trace C). These results suggest that C 2 alone does not bind effectively to HPA, and that FMNH Ϫ is the first species binding to the enzyme during catalysis.
Binding of HPA to C 2 FMNH Ϫ -After binding to C 2 to form the C 2 -FMNH Ϫ complex, the enzyme can in principle carry out the reaction through one of two paths (Fig. 10): A) C 2 -FMNH Ϫ -HPA is formed prior to reacting with oxygen, and B) C 2 -C(4a)-hydroperoxy-FMN is formed prior to binding HPA. In this section, we report investigations of the kinetics and thermodynamics of binding HPA to C 2 -FMNH Ϫ prior to reacting with oxygen. The binding kinetics of HPA to C 2 -FMNH Ϫ were investigated by double mixing stopped-flow spectrophotometry, where the first mixing added HPA to C 2 -FMNH Ϫ under anaerobic conditions to initiate the formation of the C 2 -FMNH Ϫ -HPA complex, and after various times of aging, the second mix added oxygen to form the  Fig. 7. Upon increasing the age time, formation of the C(4a)-hydroperoxy-FMN became slower as more C 2 -FMNH Ϫ -HPA complex formed. Any C 2 -FMNH Ϫ -HPA complex present reacted to form C 2 -C(4a)-hydroperoxy-FMN-HPA at ϳ54 s Ϫ1 , and the reaction was completed by ϳ90 ms. This indicates that the more complete the formation of C 2 -FMNH Ϫ -HPA, the slower was the formation of C 2 -C(4a)-hydroperoxy-FMN. This result is also consistent with our previous interpretation in Fig. 5 that C 2 -FMNH Ϫ -HPA reacts with oxygen more slowly than does C 2 -FMNH Ϫ . Fig. 7 shows that the amount of C 2 -C(4a)-hydroperoxy-FMN formed between 7 and 80 ms (the slower reaction) was maximum when age times before mixing were Ն1 s (inset in Fig. 7), indicating that binding of 2 mM HPA to C 2 -FMNH Ϫ was complete by 1 s. The apparent rate constant (k obs ) for binding of HPA to C 2 -FMNH Ϫ , calculated from the plot of the absorbance increase observed at 380 nm versus the age times after the first mixing, was 9.6 Ϯ 2 s Ϫ1 .
The thermodynamics for binding of HPA to C 2 -FMNH Ϫ prior to reacting with oxygen were investigated by double mixing stopped-flow spectrophotometry, where various concentrations of HPA were mixed with C 2 -FMNH Ϫ anaerobically in the first mix, and oxygen was added to the preformed C 2 -FMNH Ϫ -HPA complex in the second mix (Fig. 8). The age time was 10 s to ensure complete formation of the substrate complex, and the reaction was monitored at 370 and 446 nm. The final concentrations were 16 M C 2 -FMNH Ϫ , 1.03 mM oxygen, and a range of HPA concentrations. Fig. 8 shows that the kinetic traces are composed of four phases. The first phase observed is an increase in absorbance at 370 nm with an observed rate constant of 54 s Ϫ1 , and the amplitude of this phase is dependent on concentration of HPA. It is known from the single mixing experiments (Fig. 2) that the reaction of C 2 -FMNH Ϫ and 1.03 mM oxygen is fast enough to be largely completed  All reactions were carried out in 50 mM sodium phosphate buffer, pH 7.0, 4°C, in a stopped-flow spectrophotometer. The percentage of product formed was calculated from the ratio of product determined to the amount of reduced flavin used. The reagents in syringe A were specified inside the brackets, whereas syringe B contained buffer with oxygen or reduced flavin. DHPA was determined by HPLC after enzyme was removed by ultrafiltration (6).

Type of reaction Product
, and HPA (2 mM) was mixed with buffer containing 0.13 mM of oxygen. All concentrations were described as after mixing. Under these conditions, the reaction follows Path A in Fig. 10. b The solution of C 2 (25 M), HPA (2 mM), and oxygen (0.13 mM) was mixed with buffer containing 16 M of FMNH Ϫ or FADH Ϫ . All concentrations are described as after mixing. Under these conditions, the reaction follows Path B in Fig. 10. during the dead time of the stopped-flow instrument. With higher concentrations of HPA, larger fractions of the enzyme were in the form of the C 2 -FMNH Ϫ -HPA complex, resulting in smaller fractions of the enzyme reacting with oxygen during the dead time, and larger fractions reacting at 54 s Ϫ1 (Fig. 8). These amplitude changes were plotted against the concentrations of HPA, and a K d value of 180 Ϯ 3 M (referred as K d B in Fig. 10) was calculated for binding of HPA to C 2 -FMNH Ϫ (inset A in Fig. 8).
The second phase in the reactions in Fig. 8 was a small decrease in absorbance with an observed rate constant of ϳ17-22 s Ϫ1 (k 6 in Fig. 10), whereas the third phase was a small increase in absorbance with a rate constant of ϳ6 -9 s Ϫ1 (k 7 in Fig. 10). The absorbance at 370 nm decreased again in the fourth phase with the k obs values inversely dependent on the concentration of HPA used. The fourth phase was identified as the formation of the final oxidized FMN species because the traces coincided with a large increase in absorbance at 446 nm (shown in inset B). These results suggest that after formation of C 2 -C(4a)-hydroperoxy-FMN during the first phase, HPA was hydroxylated with the formation of C 2 -C(4a)-hydroxy-FMN during the second phase, similar to the results of Fig. 5. However, it is clear from this experiment that excess HPA can also bind to the enzyme to trap the C 2 -C(4a)-hydroxy-FMN-HPA species (Fig. 10) in the third phase, causing a slight increase in absorbance at 370 nm. With higher concentrations of HPA, more of this intermediate was trapped as the C 2 -C(4a)hydroxy-FMN-HPA species, so that the dehydration to form the oxidized FMN was retarded (inset B in Fig. 8). Similar trapped C(4a)hydroxyflavin-substrate species have also been observed in the oxidative half-reactions of several single component flavoprotein oxygenase enzymes (24 -26, 34 -35).
The Reaction of C 2 -C(4a)-hydroperoxy-FMN with HPA-Experiments from the previous section show that hydroxylation can occur via Path A in Fig. 10 where C 2 -FMNH Ϫ first binds to HPA and then reacts with oxygen to form C 2 -C(4a)-hydroperoxy-FMN-HPA or through Path B of Fig. 10, where the enzyme first forms C 2 -C(4a)-hydroperoxy-FMN and then binds to HPA in a following step. Therefore, the reaction of Path B was explored using a double mixing stopped-flow spectrophotometer, where the intermediate C 2 -C(4a)-hydroperoxy-FMN was generated by reacting C 2 -FMNH Ϫ with oxygen in the initial mixing; after aging for 0.1 s to fully form the C(4a)-hydroperoxy-FMN, the resultant intermediate was mixed with buffer containing various HPA concentrations. Reactions were monitored at 370 nm (Fig. 9), and the results indicate that binding to HPA (the small increase in absorbance from 2-20 ms) gave a phase with amplitudes and rates that were dependent on the concentration of HPA. Kinetic analysis showed that the observed rate constants (k obs ) of this phase were hyperbolically dependent on HPA concentrations (inset A in Fig. 9). These results are consistent with binding being a two-step process, with a rapid equilibrium in the initial step and an isomerization in the following step (Path B in Fig. 10) (36). Data were analyzed according to Equation 1, yielding a K d value for the initial binding of HPA of 0.35 Ϯ 0.03 mM (K d C , Fig. 10), and the rate constant for the subsequent isomerization of 208 Ϯ 4 s Ϫ1 (k 4 in Fig. 10).
After HPA has bound, there are three more phases in the reaction similar to those seen in the double mixing experiments described in Fig. 8. In the second phase, C 2 -C(4a)-hydroperoxy-FMN-HPA converted to C 2 -C(4a)-hydroxy-FMN-DHPA with a rate of 17-22 s Ϫ1 (k 6 in Fig. 10), and this was followed by a slight increase in absorbance during the third phase. The amplitude of the third phase is also dependent on HPA concentration. As before, the second phase is the hydroxylation step, and the third phase is the binding of HPA to C 2 -C(4a)-hydroxy-FMN, coincident with the release of DHPA. The fourth phase is the decrease in absorbance 370 nm and a large increase of absorbance 446 similar to those of Fig. 8B (data not shown). The fourth phase was interpreted as the dehydration of C 2 -C(4a)-hydroxy-FMN to form oxidized FMN, and this rate was inversely dependent on HPA concentration (as dis-   cussed in the next section and shown in inset B). When all relevant kinetic constants were considered, the results suggest that the oxygenation reaction of C 2 occurs preferentially through Path B (Fig. 10) over Path A. Therefore, the kinetic mechanism for the oxygenation reaction of C 2 -FMNH Ϫ can be described as a preferential random order mechanism as shown in Fig. 10, and the preferred path utilized will be determined by the relative concentrations of substrates. (The formal nomenclature would be Uni Bi Uni Bi mechanism with a random segment for the second and third substrates, and probably the second and third products.) To ensure that the hydroxylation reaction indeed occurred via Path B of Fig. 10, analyses were also carried out with reaction mixtures in which solutions of FMNH Ϫ (50 M) and 2 HPA (2 mM) were mixed in the stopped-flow instrument with a solution of C 2 (80 M) and HPA (2 mM) in air-saturated buffer. Under these conditions, the reaction occurred via Path B in Fig. 10, where the intermediate C(4a)-hydroperoxyflavin was formed prior to binding of HPA. Samples collected from the stopped-flow instrument were analyzed by HPLC as described previously (Table 1). Results indicated that the DHPA product yield from Path B was 82 Ϯ 3%, which is slightly greater than the value obtained from Path A as 73 Ϯ 4% (Table 1 and Fig. 5).
Inhibition of the Dehydration of C 2 -C(4a)-hydroxy-FMN by HPA-Data from experiments shown in Figs. 8 and 9 indicate that the observed rate constants of the fourth phase were inversely related to the concentration of HPA, implying that HPA binds to the enzyme to form a trapped C 2 -C(4a)-hydroxy-FMN-HPA species and impedes it from dehydrating into oxidized FMN. Consistent with this interpretation, a plot of the observed rate constants of the last phase versus HPA concentration shows an inverse dependence on HPA. The rate extrapolated to zero with increasing concentrations of HPA (inset B in Fig. 9), implying that the trapped species is a dead-end complex and is not capable of dehydrating to form the oxidized enzyme. Because the observed rate constants for dehydration are specified by two reactions, the dehydration and the rebinding of HPA to the intermediate (Fig. 10), Equation 2 was used to analyze the observed rate constants for the dehydration. When the observed rate constants of the fourth phase were fitted to Equation 2, the K d in for dissociation of HPA from the C 2 -C(4a)-hydroxy-FMN was calculated to be 41 Ϯ 1 M (K d in in Fig. 10) and the dehydration rate constant obtained from extrapolation to zero HPA was 8.3 Ϯ 2 s Ϫ1 . This dehydration rate must represent the rate of dissociation of HPA from the enzyme.
Oxygen Reaction of C 2 -FADH Ϫ in the Presence and Absence of HPA-A unique property of C 2 compared with other two-component flavindependent hydroxylases is its ability to use a variety of reduced flavin substrates. FADH Ϫ is as effective as FMNH Ϫ (6,15). Therefore, we tested whether the mechanistic details for oxygenation by C 2 with FADH Ϫ are similar to those in the reaction of C 2 with FMNH Ϫ . Experiments similar to those described in Figs. 2 and 5 were carried out, but using C 2 -FADH Ϫ instead of C 2 -FMNH Ϫ . The reaction of C 2 -FADH Ϫ with oxygen is very similar to the reaction of C 2 -FMNH Ϫ . The rate constant was 0.98 Ϯ 0.05 ϫ 10 6 M Ϫ1 s Ϫ1 for formation of C 2 -C(4a)hydroperoxy-FAD versus 1.1 Ϯ 0.1 ϫ 10 6 M Ϫ1 s Ϫ1 for the reaction with FMNH Ϫ ( Table 2, data not shown). In the presence of HPA, the C 2 -FADH Ϫ -HPA complex reacted with oxygen more slowly than in the absence of HPA, similar to the reactions with FMNH Ϫ . The rate constant for formation of the C 2 -C(4a)-hydroperoxy-FAD-HPA complex is slightly smaller than that with FMN (3.7 Ϯ 0.2 ϫ 10 4 M Ϫ1 s Ϫ1 for FAD versus 4.8 Ϯ 0.2 ϫ 10 4 M Ϫ1 s Ϫ1 for FMN (see Table 2). The spectra of C 2 -C(4a)-hydroperoxy-FAD, both in the absence and presence of HPA, calculated using the method described in the FMN experiments, are very similar to those for the C 2 -FMNH reactions (data not shown).
Double mixing experiments similar to those in Figs. 8 and 9, but using FADH Ϫ instead of FMNH Ϫ , yielded results similar to those with FMNH Ϫ . The K d values for binding of HPA to C 2 -FADH Ϫ or to C 2 -C(4a)-hydroperoxy-FAD are both similar to those with FMNH Ϫ (shown in Table 2), again emphasizing that FADH Ϫ can be used nearly as well as FMNH Ϫ by C 2 with respect to both specificity and reactivity.
Single turnover reactions of C 2 and FADH Ϫ were analyzed for the amount of hydroxylated product (Table 1) using the same protocols described in previous sessions of C 2 and FMNH Ϫ reactions. Results in Table 1 indicate that FADH Ϫ can be used by C 2 nearly as efficiently as FMNH Ϫ . The yields of DHPA obtained via Path A and B, 68 Ϯ 3 and 74 Ϯ 4%, were comparable to those for the FMNH Ϫ reaction (73 Ϯ 4 and 82 Ϯ 3%).

DISCUSSION
Our studies here have elucidated the detailed kinetic mechanism for the reactions of O 2 with reduced flavin bound to the oxygenase component (C 2 ) of HPAH from A. baumannii. The results and methods described can be used as prototypes for analyses of the two-component class of flavin-dependent oxygenases. These results clearly show that the oxygenation reaction of C 2 occurs via C(4a)-oxygenated flavin intermediates, similar to the reaction of the single component aromatic flavoprotein hydroxylases, where existence of C(4a)-flavin intermediates is well documented (1,2). It was previously found that C(4a)-hydroperoxyflavins reacted with aromatic substrates to form hydroxylated products in the reactions of p-hydroxybenzoate hydroxylase (3,33), phenol hydroxylase (37)(38)(39), melilotate hydroxylase (40), anthranilate hydroxylase (41), 2-methyl-3-hydroxypyridine-5-carboxylic acid oxygenase (24 -26), and 2-hydroxybiphenyl-3-monooxygenase (42). C(4a)-hydroperoxyflavins and C(4a)-hydroxyflavins were also detected in the oxygen reactions of HPAH from P. putida (16). The intermediates detected in the reaction of C 2 are spectrally similar to those of the enzymes mentioned. However, the less common feature of C 2 intermediates is that the C(4a)-hydroperoxyflavin and C(4a)-hydroxyflavin spectra are nearly identical; this characteristic has also been found in some mutant types of p-hydroxybenzoate hydroxylase (43). Partial resolution of spectra similar to those of C 2 -(C4a)-oxygenated intermediates was also obtained in studies of the reactions of 4-chlorophenol hydroxylase (9), the monooxygenase in the actinorhodin biosynthetic pathway (21,22), and styrene monooxygenase (13) when the enzymes were mixed with reduced flavin and limited quantities of oxygen in the absence of substrate.
Although the reaction of C 2 with O 2 is similar to the reaction of single component aromatic hydroxylases with respect to using C(4a)-hydroperoxyflavin to hydroxylate the aromatic substrate, the overall kinetic mechanism of C 2 is quite different (1-3). The first step of the reaction is binding of FMNH Ϫ to C 2 , followed by the reaction of the C 2 -FMNH Ϫ complex with oxygen to form a quite stable C 2 -C(4a)-hydroperoxyflavin. Under conditions of catalytic turnover, an aromatic substrate binds to the preformed C 2 -C(4a)-hydroperoxyflavin intermediate (Path B in Fig. 10). This contrasts with the reactions of the single component aromatic hydroxylases where the aromatic compound must be bound to the enzyme prior to reduction and reaction with oxygen. The kinetic mechanism of C 2 is remarkably similar to that for bacterial luciferases (Lux) in which the reaction of Lux-FMNH Ϫ with oxygen to form C(4a)-hydroperoxy-FMN occurs prior to binding of an aldehyde substrate (28). Although both C 2 and Lux bind more tightly to the reduced than to the oxidized flavin, the K d for the Lux-FMNH Ϫ complex is 80 nM (44), an order of magnitude smaller than that for C 2 -FMNH Ϫ (1.2 M). It is possible, however, that the C 2 -flavin complex becomes tighter after the reduced flavin is oxidized into C 2 -C(4a)-hydroperoxy FMN. The mechanism of C 2 is also similar to that for the oxygenation half-reaction of cyclohexanone monooxygenase (CHMO), where cyclohexanone binds to the enzyme after formation of the FAD-C(4a)-peroxide (29). The kinetic mechanism of C 2 has similarities to the reaction of HPAH from P. putida. It was reported that in the reaction of O 2 with the reduced flavoprotein plus the coupling protein of the P. putida HPAH, the rate of FAD-C(4a)-hydroperoxide formation is the same whether or not HPA was included in the oxygen-containing solution (1.1 ϫ 10 6 M Ϫ1 s Ϫ1 ) (16). However, as shown above, the rate for formation of the C 2 -C(4a)-hydroperoxyflavin decreased from 1.1 ϫ 10 6 M Ϫ1 s Ϫ1 to 4.8 ϫ 10 4 M Ϫ1 s Ϫ1 when HPA was pre-bound to the C 2 -FMNH Ϫ complex from A. baumannii (compare Figs. 3, 5, and 10). In the P. putida enzyme, it was also proposed that the reaction occurred via a pathway in which HPA bound to the oxygenase after the formation of C(4a)-hydroperoxy-FAD, similar to the reaction of C 2 (Path B in Fig. 10). This was consistent with the rate of HPA binding to the reduced enzyme being rather slow (16). It is possible, however, that the P. putida enzyme is actually like the A. baumannii enzyme. The reported flavoprotein of P. putida might actually be a reductase regulated by HPA, similar to that from A. baumannii (20), whereas the coupling protein could be the oxygenase that receives reduced FAD from the reductase. Experiments to test this hypothesis have never been carried out. Thus, the lack of effect of HPA on the formation of the C(4a)-hydroperoxyflavin from P. putida HPAH could be caused by HPA not binding to the oxygenase until FADH Ϫ has bound.
Reduced flavin is reactive with oxygen. Therefore, to be effective, reduced flavin-utilizing enzymes such as C 2 need to rapidly bind reduced flavin before auto-oxidation occurs. C 2 was shown in this report to bind FMNH Ϫ very rapidly (Fig. 4) with an observed rate Ն200 s Ϫ1 (compare traces B and C in Fig. 6). Such a rate corresponds to a second order rate constant of at least 10 7 M Ϫ1 s Ϫ1 , and this binding is quite tight (K d of 1.2 M under conditions studied). Therefore, the ability of C 2 to catalyze reactions without being constantly bound to the cofactor like other flavoproteins can be explained by the preferential binding of the enzyme to the reduced rather than to the oxidized flavin. Similar binding properties were also observed for the oxygenase component (HpaB) of HPAH from E. coli; HpaB binds to FADH Ϫ with a K d of 70 nM, whereas it binds to oxidized FAD with a K d of 6 M (45). Recently, a study of actinorhodin monooxygenase has shown that the oxygenase component, ActVA, binds to FMNH Ϫ with a K d of 0.4 M and to oxidized FMN with a K d of 26 M (21).
At high concentrations, HPA was found to form a dead-end complex with C 2 -C(4a)-hydroxyflavin and impede it from dehydrating to form the oxidized flavin (Figs. 8 and 9). Aromatic substrates were also found to bind to the C(4a)-hydroxy-FAD and inhibit the return to oxidized FAD in the reactions of several single component aromatic hydroxylases including phenol hydroxylase (34), 2-methyl-3-hydroxypyridine-5-carboxylic acid monooxygenase (24), and p-hydroxybenzoate hydroxylase (PHBH) (46). This type of substrate inhibition was also found in the reaction of P. putida HPAH (16). In the case of PHBH, it has been proposed that this inhibition is the natural consequence of the need for a conformational change from a solvent-free active site (for hydroxylation) to an "open" conformation for product and substrate exchange (3). Perhaps this inhibition is not important in cells, because cells are not likely to accumulate high concentrations of substrate that could cause such inhibition.
A unique property of C 2 is the ability to use a variety of reduced flavin substrates; the enzyme works well with either FADH Ϫ or FMNH Ϫ , although less efficiently with reduced riboflavin (6,15). Here we report that both C(4a)-hydroperoxy-FAD and C(4a)-hydroxy-FAD accumulated during the reaction of FADH Ϫ and C 2 with oxygen (data not shown), implying that the reaction undergoes the same pathway as that of reduced FMN. Moreover, the kinetic constants for the reaction of FADH Ϫ and FMNH Ϫ are similar ( Table 2), indicating that the reactivity of reduced FMN and FAD in each step of the C 2 reaction is nearly the same. This also implies that C 2 interacts with the reduced flavin primarily around the isoalloxazine where FAD and FMN share the same common structure. The flavin specificity of the HPAH from A. baumannii (C 2 ) comes from the reductase, which binds more specifically and tightly to FMN (6,20). This property contrasts to most other twocomponent monooxygenases, where the reductase is often less specific for the flavin whereas the oxygenase is specific for either FMNH Ϫ or for FADH Ϫ .
In conclusion, this study has elucidated the reaction mechanism of the oxygenase component (C 2 ) of the enzyme HPAH from A. baumannii. The results clearly illustrate that C(4a)-oxygenated flavin intermediates are directly involved in the hydroxylation reaction. C 2 binds to the reduced flavin (delivered from C 1 ) in the initial step, reacts with oxygen to form the C 2 -C(4a)-hydroperoxyflavin, and finally binds HPA before hydroxylation occurs. This knowledge is needed to understand catalysis by the enzymes in this two-component class. This report will be followed by a subsequent article that explains in detail the transfer of the flavin between the two protein components of the enzyme.